Bad data ... Bad decisions

Memorable phrases help people condense and retain concepts. They are the brands of ideas.

Listen to an older person today and you can still hear phrases like "Loose lips sink ships." In a marketing communications context today it should be "Damaged data dumps demand."

Damaged - or 'bad' - data can creep into your database in many ways. Sales teams can update client files without marketing noticing; external and rented lists can be inaccurate; and a company's own data collection methods can be skewed by poorly qualified prospects and suspects. It can all lead to bad decisions.

"Way too many times an organisation will continue to replicate bad data," said Jonathan Block, Senior Director-research at US marketing advisory company SiriusDecisions. In an interview with BtoB Magazine, Block said that customer relationship management software can catch duplicate names and simple mistakes, but that "it's all the other data around that contact that often continue to be dirty, which can mean that marketing doesn't have a correct view of the prospect."

According to Block, the bad data syndrome is endemic. He said between 10 percent and 25 percent of customer and prospect records include critical data errors, from incorrect demographic information to a lack of current buying disposition.

How exactly does bad data impact on-demand creation and, ultimately, conversion?

According to benchmarking statistics by SiriusDecisions, the effect is greater at the beginning of the marketing pipeline. Validating and managing data in the earliest stages of collection can lead to better lead scoring and lift conversion rates by about 25 percent between the customer inquiry stage and the point where marketing qualifies the leads.

Inappropriate marketing offers
A common problem with bad prospect information, Block said, is that marketing may continue to market to potential customers by sending offers intended for leads elsewhere in the pipeline.

The impact of bad data also affects the ability of sales to accept a lead as qualified, a process worsened by multiple databases that are not coordinated, Block said. Combining or integrated various databases well can contribute to a conversion increase of 12.5 percent.

In its analysis, SiriusDecisions examined the data cleansing practices of some 400 B2B organisations, ranging from US$20 million in revenue to more than US$1 billion.

By considering a sample prospect database of 100,000 names and a campaign response rate of 2 percent, the company estimated that an organisation with a strong commitment to data quality can produce nearly 70 percent more revenue than a company with only average data-quality procedures.

"What we've seen over the past 18 months is that marketers increasingly are focusing on the quality of their data," Block said. "As crazy as it sounds, most marketing organisations we work with feel they have enough leads and don't feel it's essential to stuff the top of the funnel. Instead, they're focusing on quality now, and the data that leads to that."

You said what?
Inaccurate survey results also contribute to bad data and can negatively impact product and marketing decisions. According to marketing research company MarketTools, companies that don't use technology to validate survey respondents can as much as triple their risk of making incorrect business decisions.

"Some people want to fill out surveys for other reasons than offering their opinions and helping companies develop their products," said Michael Conklin, Chief Methodologist at MarketTools. "Sometimes people might discover that if they say something really nice about a product every time they're asked, they stand a better chance of having someone send it to them."

Conklin said the online survey market is growing fast; more than 1 billion online business surveys were completed in 2007. But the dangers of bad data from such instruments can have far-reaching consequences.

A MarketTools' study conducted with 622 respondents, What Impact Do 'Bad Respondents' Have on Business Decisions? reveals that for a survey containing 30 percent bad respondents - those who are not real (who they purport to be), engaged (interested in really sharing an opinion), and unique (not signed up multiple times to take the same survey) - the risk of making a bad decision doubles. With 40 percent bad respondents, the risk can triple.

Moreover, the larger the survey sample, the higher the risk. The study revealed that when the number of survey takers is increased significantly - from 600 to 6,000, for example - it takes only 10 percent bad respondents to double the risk of making an inaccurate product or marketing decision.

Getting it right
Poor survey construction also can lead to bad data, and poor decision-making, according to Arthur Middleton Hughes, Vice President of the Database Marketing Institute.

"Amateur in-house survey makers might ask the same question in several different ways in different surveys," said Hughes. "So when you come back to find out the results, each survey produces different answers, and you can't combine the data over time."

Hughes said that companies can fix the most glaring database errors easily, sometimes by just examining name fields for such errors as numbers or punctuation marks, or post code fields for anything other than numbers.

"There are plenty of things you can clean up, but the one thing to remember about databases is that they always contain junk," said Hughes. "You'll never get it completely clean - ever, ever, ever. That's something to remember when you do your next direct-marketing campaign."

Get more headlines every month - Subscribe to 7 Seconds